Cost–benefit analysis of remedial interventions for the long-term unemployed

(1)

Written by ICON-INSTITUT Public Sector GmbH Michael Fertig, Márton Csillag

July – 2015

Cost–benefit analysis of

(2)

EUROPEAN COMMISSION

Directorate-General for Employment, Social Affairs and Inclusion Directorate B — Employment

Unit B.1 — Employment Strategy Contact: Gelu Calacean

E-mail: EMPL-PES-SECRETARIAT@ec.europa.eu

(3)

Directorate-General for Employment Social Affairs and Inclusion

Cost–benefit analysis of

(4)

LEGAL NOTICE

This document has been prepared for the European Commission however it reflects the views only of the authors, and the Commission cannot be held responsible for any use which may be made of the information contained therein.

More information on the European Union is available on the Internet (http://www.europa.eu). Luxembourg: Publications Office of the European Union, 2016

ISBN: 978-92-79-50580-5 doi: 10.2767/119475 © European Union, 2016

The European Network of Public Employment Services was created following a Decision of the European Parliament and Council in June 20141_{. Its objective is to reinforce PES capacity, effectiveness and efficiency.}

This activity has been developed within the work programme of the European PES Network. For further information: http://ec.europa.eu/social/PESNetwork.

This publication has received financial support from the European Union Programme for Employment and Social Innovation "EaSI" (2014-2020). For further information please consult: http://ec.europa.eu/social/easi.

1

DECISION No 573/2014/EU

Freephone number (*):

00 800 6 7 8 9 10 11

(5)

Table of contents

1. INTRODUCTION ... 6

2. POTENTIAL INTERVENTIONS FOR LTU ... 10

2.1. Empirical Evidence ... 10

2.2. Assessment of impact for case study countries ... 36

2.2.1. Assessment of impact for Germany ... 36

2.2.2. Assessment of impact for Denmark ... 36

2.2.3. Assessment of impact for Hungary ... 37

2.2.4. Assessment of impact for Italy ... 42

2.2.5. Assessment of impact for Lithuania ... 43

2.2.6. Implications for the EU-level ... 45

3. CONCLUSIONS ... 48

4. APPENDIX 1: COUNTRY-SPECIFIC BASELINE STUDIES ... 50

4.1. Germany ... 50

4.1.1. Definition and magnitude of long-term unemployment ... 50

4.1.2. Means-tested subsistence benefits for non-employed ... 50

4.1.3. Institutional responsibility for LTU and service provision ... 53

4.1.4. Institutional responsibility ... 53

4.1.5. Service provision in the Jobcenters ... 54

4.1.6. Use and costs of ALMP-measures ... 58

4.1.7. Administration costs ... 61

4.1.8. Future initiatives ... 62

4.1.9. Summary 62 4.2. A.2 Denmark ... 63

4.2.2. Insurance-based and means-tested subsistence benefits for non-employed 64 4.2.3. Institutional responsibility for LTU and service provision ... 66

4.2.7. Summary 69 4.3. A.3 Hungary ... 69

4.3.7. Summary 78 4.4. A.4 Italy ... 79

4.4.2. Unemployment benefits and means-tested subsistence benefits for non-employed ... 79

4.4.7. Summary 98 4.5. A.5 Lithuania ... 100

(6)

1. INTRODUCTION

Addressing long-term unemployment is one of the objectives of the European Network of Public Employment Services (PES), as defined in Art 2 of the Decision on enhanced

cooperation between PES. In its work programme for 2015, the Network included

activities addressing the role of PES in the area of long term unemployment including (at the request of the European Commission, EC), a working group on the integration of the long term unemployed.

The overall objective of this study is to assess the administrative costs and the cost-effectiveness of a limited number of policy intervention options for reintegration of the long term unemployed (LTU) into the labour market. To this end, research was done for a sample of five Member States which are perceived as being sufficiently diverse in models of service provision and institutional delivery models to allow conclusions to be drawn of relevance to the EU28 countries.

The specific tasks of the study are to:

 Assess costs and benefits of the current situation (the baseline) in the five countries studied. The assessment has to take into account efforts already made or already programmed.

 Analyse, through a series of case studies and through analogies with existing evaluation studies, the administrative costs per participant for integrating service provision to the long-term unemployed, for reinforced individualised services and for formalising rights and obligations in written agreements (relative to the current situation and existing policy plans).

 Provide estimates of the cost and benefits of extending existing models for provision of activation to the existing stock of long-term unemployed in the countries examined.

The selection of Germany, Denmark, Hungary, Italy and Lithuania for the country-specific baseline studies takes into account the characteristics of the national unemployment and benefits systems and the current coverage of active measures among the long term unemployed. The sample of countries allows the analysis of a sufficiently diverse institutional and legal setting as it includes countries with centralised/regionalised responsibility for services to the long-term unemployed, as well as countries with integrated/fragmented links between benefits (social and unemployment benefits), activation measures and other services.2

In general labour market oriented service provision in a wider sense (i.e. including measures of active labour market policies, ALMP) that target the long-term unemployed can have multiple effects going beyond the immediate impact on re-entering employment. PES services can help shorten the unemployment spell and reduce the

2 Note that our case study countries also cover a wide spectrum when looking at welfare regime types and government effeciveness. In European Commisssion (2015) Literature review and

identification of best practices on integrated social service delivery (Author: Ágota Scharle),

(7)

harmful consequences of unemployment by maintaining motivation for job search, improving the effectiveness of matching jobseekers with vacancies, improving employability by training programmes, providing counselling and social services and by lowering wage costs. Carefully targeted wage subsidies can reduce employers’ wage costs and thus help prevent (or delay) layoffs or encourage new hires.

At the individual level, the immediate impact of services and ALMP-measures is increased chances of reemployment. A further potential effect can be an increase in earnings, if the intervention fosters productivity (e.g. by developing skills or removing personal obstacles that impaired productivity) or eliminated employer discrimination (based on a perception of low productivity). Thus, ALMP-measures can have a potentially large effect going beyond the facilitation of job matching by tackling the underlying causes of long-term unemployment. By shortening the unemployment spell, ALMP-measures can also mitigate the harmful consequences of unemployment on poverty, social exclusion, health and particularly mental health.3

Tackling long-term unemployment tends to become increasingly difficult the longer the unemployment spell as it becomes more difficult for the jobseeker to find a job and less effort may be put into job search.4_{This may be due to several related factors, e.g. the} erosion of skills, discrimination by employers against long-term unemployed, or the ranking of job applicants by employers on the basis of their time spent in unemployment, as well as declining motivation on the part of the unemployed person.5

Macro-econometric analyses of cross-country data show that effective activation regimes can reduce benefit recipience and increase overall employment.6_{At the macro level,} labour market-oriented service provision (including ALMP-measures) may have a direct effect via several channels. Firstly, PES interventions can maintain the job search intensity of the long-term unemployed and contain wage expectations.7_{This contributes} to more flexible wage adjustments and faster economic recovery.8_{Secondly, to the} extent ALMP-measures improve skills and productivity they can increase the productive capacity of the labour force. Improving re-employment chances may not increase the overall employment rate, as participants may simply crowd out other jobseekers while labour demand remains stable. However, faster and better matching may increase overall employment and productivity.

3

A.H. Goldsmith and T. M. Diette (2012), Exploring the link between unemployment and mental

health outcomes. The SES Indicator, The SES Indicator, 5(1). K.I. Paul and K. Moser (2009), Unemployment impairs mental health: Meta-analyses. Journal of Vocational Behavior, 74(3), 264-282.

4_{Broader benefit coverage, higher unemployment benefits and longer duration of benefits tend to} increase jobseekers’ wage expectations and so tend to reduce job-search efforts and make the unemployed more “choosy” in evaluating job offers. This improves the quality of job matches but also tends to increase the duration of unemployment.

5_{For a brief review of the causes of hysterisis, see e.g. OECD Economic Outlook (2011), Chapter 5.} Persistence of High Unemployment: What Risks? What Policies?

6_{J.P. Martin (2014), Activation and Active Labour Market Policies in OECD Countries: Stylized Facts} and Evidence on their Effectiveness, IZA policy paper.

7_{Unemployed individuals may be reluctant to lower their wage expectations as the unemployment} spell lengthens, and this may be more wide-spread in economy-wide shocks when social tolerance towards long-term unemployment tends to be higher. Behavioural conditions and sanctions can counterbalance this effect.

(8)

As will become transparent in the following analysis, the available evaluation studies aimed at identifying the causal effects of interventions for long-term unemployed, only tend to focus on a sub-set of these impacts. Furthermore, it will become clear that the strength of the evidence base for different types of interventions varies considerably and that all empirical studies necessarily had to be conducted within a given institutional framework. Hence it is anything but trivial to extrapolate the findings to countries with a different institutional context even if the evidence base is strong. However, from the evidence collected it is possible to deduce a set of structured interventions which are promising with respect to cost-effectiveness and so can serve as a valuable source for further considerations regarding the handling of long-term unemployed in many EU Member States.

(9)

2. POTENTIAL INTERVENTIONS FOR LTU

The existing empirical evidence on the effectiveness of labour market oriented service provision (including ALMP-measures) tends to focus on the newly unemployed. This body of evidence suggests that job search assistance and related PES services and sanctions may be the most cost effective interventions, with direct job creation in the public sector the least effective, with wage subsidies and training measures situated somewhere in between.9_{There is also some evidence regarding particular design features of service} delivery suggesting that integrated and individualised services and measures involving the employer tend to be more effective. However the evidence is relatively weak on whether these types of intervention are equally effective for the specific sub-group of the LTU. Drawing on the theoretical and empirical literature, this report identifies five, potentially effective interventions to support the labour market integration of LTU.

The rest of this chapter (Section 2.1) starts by examining five different interventions for LTU (in a wider sense) for which reliable empirical evidence exists on their effectiveness and their efficiency. In total 46 evaluation studies referring to 12 European countries were analysed with around half of them (22) covering Germany. This is due to the fact that in 2005 in Germany the system of “unemployment benefits 2” (UB2-benefits) was introduced and has been investigated intensely since. Further empirical studies refer to Austria (1), Belgium (1), Denmark (5), France (1), Hungary (6), the Netherlands (1), Norway (1), Portugal (1), Sweden (1), Switzerland (4) and the United Kingdom (2). Hence, the studies presented here are not restricted to the case-study countries for which country-specific baseline studies can be found in the Appendix.

The following interventions will be investigated in detail:

1. Intensified co-operation (including institutional integration) between municipalities and local labour offices

2. Provision of a “standard labour-market oriented service bundle” offering individual standard support and implementation of mutual obligations, the enforcement of rights and duties

3. Using ALMP-measures in a system of “individual standard support” 4. Provision of a “high-intensity labour-market oriented service bundle”

5. Combination of “individual standard support” with specialized services for employers

In Section 2.2 an assessment of the impact of the application of these interventions is given in the case study countries together with some implications for the EU-level.

2.1. Empirical Evidence

Intervention 1: Intensified co-operation and institutional integration

Institutional co-operation

In many Member States different institutions have a specific, separate responsibility for members of the target group, or in some cases were the starting point for reforms in

(10)

recent years. A typical example of this is the situation in Germany prior to 2005 (before the introduction of the UB2-benefit system10_{). At that time a large share of unemployed} received unemployment assistance after their eligibility to unemployment benefit was exhausted. Unemployment assistance was means-tested and the duration of entitlement was typically 12 months and the replacement rate varied between 53% and 57% of previous earnings. Local labour offices were responsible for unemployment assistance. In general, all recipients had access to the services of the labour offices (including ALMP-measures) and the whole system was financed by the Federal State.

Individuals who were not eligible for unemployment assistance but had insufficient financial means received social assistance for which municipalities were responsible (including the financing of it). Able-to-work as well as non-able-to-work individuals were among the recipients of social assistance and for the former, special activation and support measures existed. Most of them were similar to standard ALMP-measures but the focus was clearly on employment measures on the secondary labour market. The social assistance system was very fragmented and the level of service provision varied considerably across municipalities. Furthermore, there were clear incentives for municipalities to bring social assistance recipients into subsidized, but not necessarily sustainable jobs. This was because after a short working period they would again be eligible for unemployment benefits and possibly also unemployment assistance (for which municipalities did not have to pay). Hence, to some extent the system functioned much like a revolving door.

Between 2001 and 2003 a pilot project (called “MoZArT”) was conducted in 30 regions (31 projects) across Germany in order to improve the co-operation between the two institutions which were responsible for social assistance (municipalities) and unemployment assistance (local labour offices) at that time. The pilot project aimed at improving co-operation between labour offices and municipalities to sustainably integrate recipients of social and unemployment assistance into the primary labour market. To achieve this, projects received financial resources to implement their approach and the specific projects (regions) had a large degree of discretion to decide on the actual method of co-operation (i.e. on target groups, on the allocation of participants, on the use of activation and support measures, on the organizational set-up etc). An evaluation study conducted on behalf of the Federal Ministry of Labour11_{identified four different} types of co-operation approaches:

 Type 1 (5 projects): Full co-operation with respect to all processes and using a holistic approach

 Selection of participants according to clear pre-defined characteristics (“rule-based” selection)

 Service provision at joint contact point  Joint case management

 Joint calculation and disbursement of benefits  All ALMP-measures available for all participants  Employees worked spatially close to one another

10_{For details on the UB2-benefit system in Germany see baseline study in the Appendix.}

(11)

 Type 2 (12 projects): High co-operation

 Central difference to type 1: no joint calculation and disbursement of benefits

 Type 3 (7 projects): Medium or low co-operation

 Joint contact points served more as coordinating unit between both institutions than as a service point for participants

 Service provision remained in original institutions  Focus on provision of ALMP-measures

 Type 4 (7 projects): Case-based co-operation

 Co-operation based on individual cases (i.e. benefit recipients)  Cases were selected in an ad-hoc process without a binding rule

 No joint contact point but only interface coordinator between both institutions

 Primary aim: joint assessment of cases to find suitable ALMP-measures The four types represent two fundamentally different models of institutional co-operation: The first two types can be characterized as a (predominantly or fully) cooperative model with a holistic approach to participants offering some form of a common contact/access point or one-stop-shop in which either: (i) both benefit disbursement and labour market oriented service provision are organized together, or (ii) only the latter is conducted. This model focused more on guidance and counselling and, therefore, was rather personnel intensive. By contrast, the last two types can be characterized as (mostly or purely) case-based models with a strong focus on active measures of labour market policy.

Evaluation results regarding effectiveness: The evaluation study estimated the effectiveness of the projects disaggregated by type using survey data for participants and non-participants. Results suggest that the projects in types 1 and 2 were the most effective with respect to a sustainable integration (seven months and more) of participants into the primary labour market (+14% relative to the comparison group). Participants in projects of type 3 also exhibited significantly higher sustainable integration rates than comparable non-participants (+9%). By contrast the sustainable integration effect of type 4 projects was significantly and also substantially negative (-41%).

Evaluation results regarding costs: The evaluation study reported the additional costs per participant of the pilot project disaggregated by type and form of costs. Interestingly, the fully co-operating type 1 entailed the lowest costs per participant, although it was the most personnel intensive type of operation. By contrast, the loosest form of co-operation (case-based type 4) incurred the highest costs and showed the highest share of active measures.

Cost-benefit-balance: For cost-benefit-analyses the above mentioned costs were

(12)

Institutional integration

Based on the predominantly positive experiences with the “MoZArT”-projects, the UB2-system was introduced in 2005. With this introduction the legislator (German Bundestag) included a legal mandate to evaluate the so-called “model of service provision” for UB2-benefit recipients. This basically implied that the performance of (i) the municipal Jobcenters had to be compared to that of (ii) the co-operative Jobcenters. In model (i) responsibility for benefit calculation and disbursement, as well as for the provision of labour market-oriented services, is handed over to municipalities and local labour offices are not in any way involved in these processes. In contrast, in model (ii) local labour offices and municipalities co-operate in these processes and formed a new legal entity called “ARGE”. Between 2006 and 2008 a large-scale evaluation study involving many academic institutes was conducted. The evaluation examined, among other things, the causal impact of the “model of service provision” on the individual as well as the regional level using administrative and survey data. The results were summarized in a report for Parliament (see Deutscher Bundestag 200812_).

The most important findings were as follows:

1. Co-operative Jobcenters performed better than municipal Jobcenters with respect to all important outcome indicators on the individual as well as on the regional level during the one year observation period (i.e. the year 2007) of the outcomes. 2. Specifically controlling for the impact of other factors, the “ARGE” had significantly

higher integration rates into the primary labour market (between two and four percentage points) and significantly lower rates of individuals receiving UB2-benefits than the municipal Jobcenters (between three and four percentage points). Hence, the causal effect of service provision by municipal compared to co-operative Jobcenters was significantly negative.

3. Furthermore, on the level of the economy as a whole, the savings incurred by fully implementing the ARGE-model instead of the municipal model were estimated to amount to 3.3 billion € per year. Given that the number of “able-to-work persons in need” at that time was around 5 million, this is equivalent to 660 € per person or foregone benefit savings of around 10%.

These findings suggest that attempts to integrate institutions for the provision of services for the target group are more promising with respect to effectiveness and also efficiency if they combine the expertise of existing institutions instead of handing over the responsibility to only one of them.

A second example of the introduction of an integrated institutional arrangement for the delivery of public employment services and social security benefit payments for all working-age individuals (both unemployed and inactive) is the introduction of Jobcentre Plus in the United Kingdom between 2001 and 2006. Prior to this reform, the Employment Service was responsible for providing job-search related support and activation services to claimants of the Jobseeker’s Allowance (JSA) (the main benefit for

(13)

unemployed persons), while the Benefits Agency was responsible for the administration of benefit claims and benefit payments as well as running the Social Security Offices (which provided income support for inactive individuals). The services provided by the Jobcentre Plus offices range from benefit claims processing through work-focused interviews, enhanced job brokering services, referral of various activation programmes and follow-up activities.13

The evidence from the evaluation14_{, which exploits the incremental implementation of the} Jobcentre Plus between 2001 and 2006 in different local labour markets, support the view that integrated service delivery is both effective and cost-efficient. The authors estimate that the (quarterly) flows from JSA to jobs increased by 3-4% in the long run as a result of service integration.15_{The study estimated that the stock of JSA claimants} would have decreased solely due to the introduction of the Jobcentre Plus by 10% in the longer run of four years after the national rollout has been completed. This positive outcome is estimated to lead to substantial savings over a fifteen-year period: direct gains from welfare benefit savings are around 4 billion GBP; the indirect gains through increased tax payments are around 2 billion GBP; while the the roll-out costs of Jobcentre Plus were 1.9 billion GBP (though it is worth noting that the largest part of this outlay, 0.8 billion GBP was devoted to investment into refurbishments and acquisitions)16_{. When considering the impact of this new model of service delivery it} should be remembered that it entailed (besides institutional integration) the introduction of performance targets and monitoring, a modernisation of IT systems, and enhanced job brokering services, plus that it was complementary to the new welfare-to-work policies that were implemented previously.

By contrast to the findings from Germany and UK, a recent evaluation17_{of the first phase} of establishing one-stop shops in the “Norwegian Employment and Welfare Administration” reform yielded more mixed results. This reform, which was implemented from 2006 through to 2011, entailed the merger of two existing public bodies, the employment service administration and the national insurance administration, as well as a close co-operation with the municipality-based social welfare services. The evaluation, which took advantage of the step-wise implementation of the reforms over 2006-2010, found that establishment of one-stop-shops (NAV) led in the initial two years after implementation to a small negative effect on the job-finding success of LTU. Although

13 _{In the original system, income support claimants had no access to employment services, which} substantially reduced their potential to look for and find employment. The upkeep of the system was relatively expensive and therefore, the Jobcentre Plus initiative was also strongly motivated by cost-cutting considerations.

14 _{R. Riley, H. Bewley, S. Kirby, A. Rincon-Aznar and A. George (2011), The introduction of} Jobcentre Plus: An evaluation of labour market impacts. Department for Work and Pensions Research Report No. 781.

15 _{These results were similar both for short-term JSA claimants, and those who have been on} benefits for more than 6 months. The effects for lone parents were even larger, while those for disabled welfare recipients were mixed.

16 _{National Audit Office, UK. (2008), Department for Work and Pensions: The roll-out of the} Jobcentre Plus Office network. London: TSO.

(14)

these negative consequences disappeared from the third year after implementation, the authors were unable to discern significantly positive effects.18

Finally, some practical issues with respect to the implementation of both institutional co-operation as well as institutional integration have to be considered which can be seen as

implementation risks. The existing experiences indicate that, firstly, for both options an

initial financial outlay (once-only knock-on financing) is necessary to enable the new structure (either cooperative or integrated) to start its work. It is typically used to finance compatible IT systems, to equip the joint contact/access point, and for the training of personnel, etc. Naturally, the amount required will be lower the higher the degree of institutional cooperation prior to the introduction of the new structure was and the less fundamental the organisational change associated with its introduction is. In Germany the initial financing in the first year (2005) of the introduction of the UB2-benefit system, which can be seen as a rather fundamental organisational change, amounted to 175€ per household of the target group. The costs of the introduction of Jobcentre Plus in the UK were around 275€ per benefit recipient (excluding infrastructure investment costs). Experiences from Hungary collected in the context of the modernisation of the PES suggest that initial financing of about 300 € per person can be necessary. It seems reasonable to assume therefore that currently a once-only set-up costs of between 200 € per household and 300 € per person of the target group constitutes an upper limit and that, with less complex reorganisations, on the set-up costs may be lower.19

Secondly, possible legal obstacles may need to be considered. Depending on the data protection regulations, practical problems with respect to the consolidation of information from different systems can occur. Similarly depending on public services law and the co-determination rights of the work councils, labour law-related problems can also occur. This is especially relevant for cases in which public employees from different institutions are supposed to work in a newly formed organisation and concerns aspects such as differences in remuneration or the authorisation to issue directives.20_{Furthermore, since} a precise forecast of the necessary personnel for the new structure is difficult to make, it is likely that during the start-up years a considerable share of employees with temporary work contracts will emerge. Depending on the specific regulations of a Member State regarding such contracts, this might also entail some practical problems.

18_{This could be due to the fact that most users still demand specialized case worker competence,} inadequate planning of IT infrastructure, large training requirements of staff, increased number of users during the 2008 financial crisis, and the build-up of new specialized teams in 2008 that took competent workers away from the day-to-day operation of the new NAV organization. 19_{It also has to be noted that these initial relatively large outlays can only be recovered in the form}

of benefits savings and increased taxes over a relatively long period, as the experience of Jobcentre Plus showed.

(15)

Thirdly, it has to be kept in mind that the introduction of the new structure can be associated with serious acceptance problems among members of the target group, employees in the existing institutions (“clash of organizational cultures”) and society as a whole. These problems might be especially pronounced if the new structure is primarily perceived as a forced means to save public money instead of a way to support the target group towards improved living standards. For this reason, an encompassing communication strategy with informing and advertising elements prior to the introduction is recommendable, though will be associated with some additional costs.

(16)

Intervention 2: Provision of a “standard labour market oriented service bundle” (“individual standard support”) and implementation of mutual obligations (“rights and duties”)

This service bundle includes:

(i) individual (qualified) contact person for each recipient, (ii) assessment of employment potential (profiling), (iii) agreement on individual action plan (IAP),

(iv) access to the full spectrum of ALMP-measures, in the sense that the target group is eligible for such measures, but contact persons have discretion to decide whether a person is allowed to participate in a measure or not,

(v) provision of job offers and

(vi) regular follow-up of profiling-IAP process (contact interval of three to six months)

(vii) with caseloads between 1:150 and 1:250.

This service bundle can be considered as a minimum requirement for any “individual support” to LTU. It contains an intimately related system of services that necessarily belong together. Any agreement on an IAP has to be preceded by a thorough assessment of an individuals’ employment potential (i.e. a profiling) and has to be followed up regularly. Furthermore, within this process it is necessary to have the possibility to refer jobseekers to ALMP-measures if the IAP suggests that this is promising. In other words, implementing only specific elements of this bundle contradicts the idea of the integrated individual support aimed for within the intervention under discussion. Furthermore, it is worth noting that the additional costs of each element are rather small. Once the groundwork for a profiling system is set up, the additional costs of an agreement on an IAP are small since the person is in contact with his/her counsellor and the elements of the IAP should be a straightforward outcome of the profiling results. The same argument then applies to the provision of job offers or the referral to ALMP-measures. Finally, it is worth noting that an IAP (in written form) provides a legal basis for the implementation of sanctions (for details see below).

(17)

To what extent these costs are additional in the sense that they were incurred on top of the costs for the two systems (unemployment and social assistance) which existed prior to the introduction of the UB2-system, is very difficult to assess because the costs of the old systems are unknown. However, it is axiomatic that the administrative costs for the calculation and disbursement of benefits in the new system do not exceed those of the old one because the task also had to be performed in the old systems. On the contrary, it is plausible that through a unified system of benefit calculation and disbursement, synergetic effects can be achieved because, for example, information on the household context of a benefit applicant no longer has to be collected twice. However, to remain on the conservative side in the estimates provided, it is assumed that the additional costs per benefit recipient of this task are zero.

By contrast, the costs of labour market-oriented service provision are in all likelihood higher in the UB2-system than in its predecessor systems because the coverage of service provision is higher. Unfortunately, there is no information on personnel or caseloads in the old systems so the only option is to rely on possible scenarios. Assuming that factual caseloads were 50% higher in the old systems than in the UB2-system, the administrative costs of the old systems can be estimated to be two-thirds of the new system (i.e. around 1.7 billion €). Under the assumption that caseloads in the old systems were twice as high as in the new system, the administrative costs of the old systems can be estimated to amount to 1.3 billion €. In these two scenarios the additional administrative costs of labour market-oriented service provision in the UB2-system add up to 0.9 to 1.3 billion € per year.

In 2013 total payments to the average "household in need" were 10,132 € (see baseline study for Germany). Thus, to compensate for these additional costs by benefit savings only (i.e. ignoring tax and social security contribution revenues) between 89,000 and 128,000 UB2-benefit-recipients have to leave the UB2-system for one year. According to official statistics around 1.8 million “able-to-work persons in need” left the benefit records during 201321_{. Around 75% of them did not re-enter the system again within} three months. Thus, around 450.000 people left the system for up to three months and 1.35 million people did not receive UB2-benefits for at least four months. Assuming that individuals in the first group stay out of the UB2-system for 1.5 months on average and individuals in the second group for eight months on average, the number of persons who left the UB2-system for one year is equivalent to 956,250. Therefore to achieve the 89,000 to 128,000 individuals necessary to compensate for additional administrative costs of the UB2-system by savings in benefits alone, between 9% and 13% of these outflows from the system have to be causally attributable to the UB2-system itself. If those who have left the system also pay taxes and social security contributions, even lower shares are required to arrive at cost-neutrality. Clearly this is a strong causal impact, but it appears to be realistic.

Some evidence on the costs and gains from an incremental move towards providing this type service bundle can be obtained from an evaluation study of the modernisation of the Hungarian PES between 2004 and 2008 (financed by HRDOP 1.2 measure). This project involved five elements: the introduction of the new service model (essentially allocating

(18)

services and measures to clients based on profiling), the remodelling and renovation of offices (introduction of self-service points etc.), the introduction of a quality assurance system, the training of PES staff and the introduction of an integrated IT system. The first four of these elements were introduced in the 71 participating local offices (out of a total of 158 offices of the HPES), while the last one was of course introduced across the whole administration. The modernisation is estimated22_{to have increased the outflow} rate to jobs in the primary labour market by 6% to 10%, meaning that clients’ unemployment spells shortened by 1.5 to 2 months in participating local offices. The estimated monthly costs of the programme per participant were about 300 € and the benefit-to-cost ratio was about 2 to 2.5:1.23

A prerequisite to more individualised support is the existence of a well-designed profiling (and targeting) system. While the benefits of having such a system are difficult to evaluate empirically, a handful of studies24_{indicate that caseworker allocation of} unemployed to ALMP-measures tend to do no better than random allocation, and that a targeting system can significantly increase the re-employment rates of unemployed.25 While it is well known that profiling systems cannot be implemented without accompanying incentive and organisational changes, it is worth considering the costs (and benefits) of setting up such a system not only for preventing long-term unemployment, but also to formulate (econometric analysis based) guidelines for caseworkers on the optimal allocation of individuals to programmes.26

Rights and duties: the use of sanctions in “individual standard support”

One further important aspect of a system of individual support is the implementation of mutual obligations (“rights and duties”, i.e. the possibility to cut benefits if jobseekers do not fulfil their obligations). Such cuts are commonly known as sanctions and are widely-used. When considering the effect of benefit cuts, differentiation has to be made between (i) the ex-ante effect, i.e. the possibility that unemployed persons’ behaviour will be monitored (in other words, the behavioural response to a move from a system with no

22_{Zs. Cseres-Gergely (2011), Greasing the wheels of the labour market? Impact analysis of the} modernization of the Public Employment Service of Hungary. The Hungarian Labour Market

2011. Budapest: IE-HAS, p. 82-95.

23_{Please note that this estimation pertains to all registered unemployed clients of the PES. For} calculating costs and gains, it is useful to note that the average welfare benefit at the time was roughly 95 € per month and the taxes and contributions from a minimum wage job amounted to about 130 € per month.

24_{These are: M. Lechner and J. Smith (2007), What is the value added by case workers?, Labour} Economics, vol. 14, p. 135-151; J. Staghoj, M. Svarer and M. Rosholm (2010), Choosing the Best Training Programme: Is there a Case for Statistical Treatment Rules? Oxford Bulletin of Economics and Statistics, vol. 72,p. 172-201; M. Huber, M. Lechner, C. Wunsch and T. Walter (2011), Do German Welfare-To-Work Programmes Reduce Welfare and Increase Work?, German Economic Review, Vol. 12, p. 182-204.

25 _{Lechner and Smith (2007) indicate that the employment rate one year after entry into} unemployment could be raised by 8 percentage points in Switzerland, Staghoj et al. (2010) indicate that Danish jobseekers‘ duration of unemployment could be reduced by about 8 weeks, Huber et. al. (2011) estimate that the optimal allocation to ALMPs would reduce welfare receipt by 9 percentage points; these amount to about a 12-18% improvement in outcomes.

(19)

sanctions to one with sanctions), and (ii) the ex post effect, i.e. the reaction of unemployed to the actual imposition of a sanction.

While there is some evidence that more intensive monitoring of job search activity leads to an increase in re-employment27_{and that activation programmes work through} ´threat´ effects28_{, the most clear-cut indication of anticipation effects come from a} recent study in Belgium. This study29_{investigated, in the context of a reform where job} search requirements were introduced for younger long-term unemployed, the impact of the notification that job search activities will be controlled at a later date. This notification was performed prior to the start of the monitoring. The study found that the transition to employment was 22-28% higher among individuals affected by the reform (and notified) than among those unaffected. This study provides support to the hypothesis that a relatively low-cost initiative (i.e. the more effective enforcement of jobseekers’ obligations) can lead to important gains in flows out of unemployment.30

Two studies from Switzerland provide additional evidence on the favourable effect on the ‘threat’ of sanctions. According to the first paper31_{, caseworkers insisting on unemployed} persons’ duties are found to increase their clients’ employment rate more than their more-cooperative caseworkers, and without pushing them into unstable jobs. The second study32_{clarifies this picture showing that raising job search requirements (the number of} job applications to be submitted) does indeed raise re-employment probability, but this comes at the cost of accepting lower wages and this strategy only seems to work in relatively favourable labour market conditions.

In Germany “able-to-work persons in need” (i.e. persons receiving UB2-benefits) are obliged to regularly visit Jobcenters, to document job search activities, to accept job offers, to accept referrals to ALMP-measures and to sign an IAP which has to be up-dated at least every six months. Those who do not fulfil these obligations can be sanctioned. Sanctions are regulated by law and vary between 30% and 100%. A number of studies empirically examine the effects of sanctions33_{using different datasets and methods.}

27_{See for example: J. Micklewirght and Gy Nagy (2010), The effect of monitoring unemployment} insurance recipients on unemployment duration: Evidence from a field experiment. Labour Economics, 2010, vol. 17, p. 180-187.

28_{See for example: B. K. Graversen and J. van Ours (2011), An Activation Program as a Stick to} Job Finding. Labour, 2011, col. 25, 167 – 181.

29_{Cockx, B. and M. Dejemeppe (2012), Monitoring job search effort: An evaluation based on a} regression discontinuity design. Labour Economics, vo. 19 , p.729–737

30_{It is less clear whether transitions due to threat effects are to lower the quality of jobs (e.g.} regarding stability), and in this sense, whether the gains are long-term.

31_{S. Behncke, M. Frölich and M. Lechner (2010), Unemployed and Their Caseworkers: Should They} Be Friends or Foes? Journal of Royal Statistical Society, Series A, 173 (1), 67-92

32 _{P. Arni and A. Schiprowski (2015), The Effects of Binding and Non-Binding Job Search} Requirements. IZA DP No. 8951.

33_{B. Boockmann, S. Thomsen and T. Walter (2009), Intensifying the Use of Benefit Sanctions? An} Effective Tool to Shorten Welfare Receipt and Speed up Transitions to Employment? Zentrum für Europäische Wirtschaftsforschung (ZEW), Discussion Paper 09-072. K. Hillmann and I. Hohenleitner (2012), Impact of Benefit Sanctions on Unemployment Outflow – Evidence from German Survey Data. Hamburgisches Welt-Wirtschafts-Institut (HWWI), Research Paper 129. J. Schneider (2008), The effect of unemployment benefit II sanctions on reservation wages. IAB-Discussion Paper Nr. 19. J. Schneider (2010), Impacts of Benefit Sanctions on Reservation Wages, Search Effort and Re-employment. In: Activation of Welfare Recipients: Impacts of Selected Policies on Reservation Wages, Search Effort, Re-employment and Health.

(20)

Despite this heterogeneity regarding data sources and identification strategies, all studies conclude that sanctions increase the probability of employment on the primary labour market for those UB2-benefit recipients who experienced them.34_{What remains} controversial is the magnitude of the effects of sanctions and whether there is only an effect of factual sanctioning (i.e. the factual use of the possibility to cut benefits) or also a ´threatening´ effect (i.e. an effect of the pure possibility of cutting benefits). With respect to the latter, a survey35_{among UB2-benefit recipients in 2013 suggests that} there is indeed a ´threatening´ effect which is, however, difficult to quantify.

The factual sanctioning experience in the UB2-system in Germany during the last years indicates that only some 3% of all “able-to-work persons in need” experienced a benefit cut, of on average 20-25% per year (see also baseline study for Germany). Thus, the rate of sanctions and the cuts are rather low. However, even in these cases benefit savings of around 200 million € per year were achieved. Thus, the direct counter-financing contribution of even a restrained use of sanctions with moderate rates can add up to 20% of the additional costs of running a system of individual support.

Intervention 3: Using ALMP-measures in a system of “individual standard support”

The country-specific baseline studies revealed that ALMP-measures are widely used for LTU, sometimes on a large scale. As will become clear below, the available evidence from several evaluation studies suggests that only some programmes are promising with respect to the employment prospects of participants. In general, the decisive prerequisite for them to be promising is an allocation process which is careful, if not meticulous. The decisive prerequisite in turn for such a meticulous allocation is a clear understanding of the particular problems and needs of each and every member of the target group. To this end an appropriate counselling process has to be established first.

In the following the effectiveness and efficiency of several groups of ALMP-measures are discussed which can in principle be used in a system of “individual standard support”. These measures are:

(i) employment on secondary labour market, (ii) short-term training measures,

(iii) self-employment subsidies, (iv) re-employment bonuses and (v) complex programmes.

The use of wage subsidies will be analysed in intervention 5 below since wage subsidies can be one approach to address employers.

Employment Effects of an Intensified Use of Benefit Sanctions. In: Walter, T. (2012):

Germany's 2005 Welfare Reform. Evaluating Key Characteristics with a Focus on Immigrants. ZEW Economic Studies, 46, 51-72.

34_{Similar results were found for sanctions among welfare recipients in the Netherlands where, after} a benefit penalty has been imposed, the transition rate out of welfare increased by 35-50%. See: B. van der Klaauw and J. van Ours (2013), Carrot and stick: How re-employment bonuses and benefit sanctions affect exit rates from welfare. Journal of Applied Econometrics, vol. 28: 275–296

(21)

Ad (i) medium-term employment on secondary labour market

In Germany the so-called “1€-Jobs” have been a widely used ALMP-measure which is explicitly directed and exclusively available for recipients of UB2-benefits. All these jobs are in the secondary labour market, have to be “additional", "in the public interest", and "non-competitive" with other employers. The target group consists of benefit recipients who are the “hardest to place” on the primary labour market due to multiple problems. The measure aims at improving their employability and at contributing to their employment on the primary labour market. Furthermore, it is used to check their compliance with behavioural conditions (participation in these jobs is not voluntary). On average around 340 € was spent per participant in 2013 and participants spent 4.5 months in such a job. Thus, total expenditures per person (without benefit payments) for a typical participant in such a job amounts to approximately 1,530 €. This spending includes the “wage” for participant of around 1.5-2 € per hour of work and a lump sum for the provider of these jobs. The average stock of participants in “1€-Jobs” in 2013 was 111,428 individuals.36

For these “1€-Jobs” several evaluation studies exist37_{. In general, for the average} participant these studies find a non-positive (i.e. zero or negative) causal effect of participation in such jobs on employment prospects in the primary labour market and on leaving the benefit system. Only for selected and comparatively small sub-groups a significant increase in the employment prospects can be established in the medium-run. By contrast, for some other sub-groups (especially young and rather experienced participants) a lasting negative impact on the chances to find a job on the primary labour market can be observed.

These zero or even negative employment effects are due to the fact that the allocation into these jobs is mostly not optimal. This means that a large share of the participants consists of anything but the “hardest to place” individuals. This misallocation results in sometimes large lock-in effects during participation and no positive or even lasting negative effects on employment chances on the primary labour market after the end of the measure.

The evidence about “public works” programmes (which typically last 5 months) in Hungary point to similar negative phenomena.38_{A large majority of the programme} participants consists of disadvantaged persons (low-educated, Roma background, living in micro-regions with low employment rates). The programme has a negative effect for

36_{Compared to the preceding years this was a rather small number. In the years 2006 to 2010 the} average stock of participants amounted to more than 300,000 individuals and the “1€-Jobs” were the single measure with the largest number of participants. In 2013 more participants were only counted in the so-called “training measures” (see below).

37_{See e.g. K. Hohmeyer and J. Wolff (2012), A fistful of Euros: Is the German one-euro job} workfare scheme effective for participants? International Journal of Social Welfare, Vol. 21, No. 2, S. 174-185. S. Koch and M. Fertig (2012), Evaluation von Arbeitsgelegenheiten in der Mehraufwandsvariante im Jobcenter München. IAB-Forschungsbericht, 01/2012, Nürnberg. IAB/ISG (2011), Evaluation von Beschäftigung schaffenden Maßnahmen nach § 16d und § 16e SGB II in Hamburg. Endbericht, Download: http://doku.iab.de/externe/2011/k110725301.pdf. 38_{Zs. Cseres-Gergely and Gy. Molnár (2014), Közmunka, segélyezés, elsődleges és másodlagos}

munkaerőpiac [Public works, welfare benefits, primary and secondary labour market.] In: Társadalmi Riport 2014, Budapest: TÁRKI, p. 204-225.; R. Csoba and Z. E. Nagy (2011), The evaluation of training, wage subsidy and public works programs in Hungary. In: The Hungarian

(22)

the re-employment probability on the primary labour market, especially for the low-skilled.39_{A similarly important phenomenon is that persons who participated previously} in public works programmes have a high chance of participating again and that repeated participation has a particularly detrimental effect on re-employment prospects. In other words, programme participants are often characterised by recurring movements between welfare benefits and public works.

Finally, in the ESF-funded programme “Kommunal-Kombi” (“KoKo”, January 2008 to December 2012) in Germany jobs on the secondary labour market were subsidized for up to three years. Predominantly communities, but also non-profit organisations in regions with relatively high unemployment rates of 15% (later reduced to 10%) could apply for subsidies for workplaces which had to be “additional", "in the public interest", and "non-competitive" with other employers. Employers were not obliged to provide special support or coaching for participants. The target group consisted of LTU who had to have been unemployed for at least two years. Altogether 15,825 jobs were subsidized for which in total 632 million € was spent (i.e. almost 40,000 € per job).

The programme was evaluated40_{with the following central finding that it had a} significantly negative impact on participants’ probability to be employed on the primary labour market and their stability of employment even in the long-run. Participants in “KoKo” experience slightly better employment prospects on the primary labour market, but only 45 months after entry into the programme. Furthermore, results of different cost-benefit-analysis scenarios suggest that the programme induced net total costs between roughly 120 million € and 240 million €. Relative to the 15,825 subsidized jobs this is equivalent to total net costs of between about 7.600 € and 15,200 € per subsidized job on the secondary labour market.

Ad (ii) short-term training measures

One of the most often used ALMP-measures in the German UB2-system is short-term training which typically comprises courses such as job application training, ability diagnosis, hardware and software training, preparation for self-employment, etc. Such training measures can also be implemented as internships in private companies. On average around 21% of all participants in ALMP-measures for UB2-benefit recipients between 2009 and 2013 can be found in such measures. In 2013 the average stock of participants in short-term training measures was 136,580, of which only slightly more than 6,000 were implemented in private companies. These measures are relatively cheap not only because average spending per participant amounts to about 400-450 € per month, but also because the average duration of participation is no longer than 2.5 months (see baseline study for Germany). Thus, for the typical participant expenditures per head for such a measure were around 1,000-1,125 €.

39 _{Public works can be organised and implemented by a variety of organisations, the largest} proportion of persons on public works participate in programmes organised by municipalities, which show the worst results in terms of exit to the primary labour market, and are also characterised by a large variation in the quality of the programmes.

40_{See IAW/ISG (2013), Programmbegleitende und abschließende Evaluation des} Bundesprogramms Kommunal-Kombi. Endbericht. Download:

(23)

Several evaluation studies examined the impact of these training measures on the labour market prospects of participants41_{. The following important conclusions can be drawn:}

 Training measures in private companies have a significant and substantial causal impact on the employment prospects of participants on the primary labour market in the short as well as in the long-run. For instance, the probability to be employed on the primary labour market 28 months after participation in such a measure is higher by 21 percentage points for participants than comparable non-participants.

 Training measures which are organized in courses (i.e. the majority of these measures) exhibit more mixed results. Depending on the content, significantly positive employment effects of up to four percentage points can be observed. However, negative and insignificant employment effects have also been found, especially for job application training.

 In general, short-term training measures can be perceived as one of the most promising group of ALMP-measures for UB2-benefit recipients with respect to employment on the primary labour market. The positive effects are especially pronounced for young participants (under 25 years of age) for which short-term training measures are much more effective than any other instrument of ALMP.

 Moreover, given the relatively low costs per head and the relatively short duration of participation, these measures are in all likelihood42_{also efficient. In order to} compensate the typical expenditures per head for such a measure of 1,000-1,125 € the average “household in need” has to be off the benefit records for only about 1.5 months.

Ad (iii) self-employment subsidies

Together with the UB2-system in 2005 a new form of subsidy for benefit recipients who wanted to become self-employed was introduced (the so-called “Einstiegsgeld”). The average stock of UB2-benefit recipients being supported by this subsidy declined fairly steadily from almost 20,000 in 2007 to around 4,300 in 2013. In 2013 the average expenditure per participant was 239 € for an average duration on the programme of 9.4 months. Hence, total spending for a typical participant amounted to around 2,247 €. Compared to the previous years, the expenditures per head remained more or less the same.

For this self-employment subsidy just one evaluation study43_{exists. In it the authors} studied UB2-benefit recipients who entered the programme from February to April 2005 and compared them to similar non-participants to estimate the impact of the programme on two outcomes “neither being registered as unemployed nor as a job-seeker” and “no

41_{E. Jozwiak and J. Wolff. (2007), Wirkungsanalyse: Kurz und bündig – Trainingsmaßnahmen im} SGB II, IAB-Kurzbericht, 24/2007, Nürnberg. Kopf, E. (2009), Short-term training variety for welfare recipients: the effects of different training types, IAB-Discussion Paper Nr. 17/2009, Nürnberg. J. Wolff and E. Jozwiak (2007): Does Short-Term Training Activate Means-Tested Unemployment Benefit Recipients in Germany?, IAB-Discussion Paper No. 29/2007, Nürnberg. 42_{Unfortunately, no cost-benefit-analysis for short-term training measures exists.}

(24)

receipt of UB2-benefit”. Estimation results show that by the time almost none of the participants no longer received the start-up subsidy, the self-employment subsidy considerably reduced the proportion of registered jobseekers and of means-tested benefit recipients among participants. Concerning the latter, the difference between participants and comparable non-participants reached more than 15 percentage points two years after entry into the programme. Although no cost-benefit-analysis was conducted, this result clearly suggests that the programme is not only effective but also efficient since direct programme costs are comparatively low.

Ad (iv) re-employment bonuses

In many European countries a large proportion of income when moving from unemployment or social assistance to employment is ‘taxed away’. Hence it is worth considering policies that ‘make work pay’. Positive results on re-employment bonuses have recently been provided by the UK Employment Retention and Advancement Demonstration. In this randomised control trial (RCT) among long-term unemployed (above age 25) receiving the means-tested Jobseekers’ Allowance, those in the treatment group received a retention bonus of, up to six payments of 400 GBP for each quarter when participants worked 30 or more hours per week for 13 out of 17 weeks. They also received coaching sessions on how to effectively advance in the workplace.44 The evaluation of this RCT45_{showed that there was a small positive impact on} employment rates of participants, which was still present after the period when re-employment bonuses ceased. As a result the programme was cost effective, with a 4:1 benefit to cost ratio. Even more positive results have been found recently in a pilot study of income tax credits to lone parents in Denmark. During this pilot, upon re-employment long-term non-employed lone parents with young children could receive a tax credit of a maximum of 600 DKr per month. An evaluation46_{revealed that during the two year pilot,} the target group spent about six weeks more in employment than the comparison group (lone parents with older children), and that the intervention was cost efficient, with a benefit-to-cost ratio of 10:1.

Ad (v) complex programmes

Recent evidence on setting up a combined bundle of services and measures comes from Portugal, where in early 2012 the ‘Convocatórias’ activation programme was launched. The programme, targeted at jobseekers who have been unemployed for at least six months or older than 45, includes mandatory participation in intensified meetings with PES counsellors, followed by targeted ALMP measures. This large-scale programme (with 80,000 entrants during 2012) signified an important shift towards activation in Portugal, in a context where both the unemployment rate and the proportion of long-term

44_{Note that the programme also comprised a training bonus, meaning a tuition payment of up to} 1,000 GBP if training was undertaken while working. However, the ERA had no impact on training rates.

45_{R. Hendra, J:A. Riccio, R. Dorsett, D.H. Greenberg, G. Knight, J. Phillips, P.K. Robins et al.} (2011), Breaking the Low-Pay, No-Pay Cycle: Final Evidence from the UK Employment Retention and Advancement (ERA) Demonstration. Department for Work and Pensions Research Report No. 765.

(25)

unemployed was high and rising. An evaluation47_{of the impact of the programme on} long-term unemployed below age 45, found that the re-employment probabilities of the affected individuals increased by at least 50%. While not performing detailed cost-benefit calculations, the authors indicate that the programme incurred a saving of around 240 million € due to reduced unemployment benefit payments in the first year of the programme. However, more information is needed to evaluate the cost efficiency, since: (i) the costs of launching the programme are not provided; (ii) the authors do not take into account increased social security and tax revenues, nor potential displacement effects; (iii) it is not straightforward to extrapolate results to those who have been unemployed for over one year.

The success of more individualised (complex) services and measures to help long-term unemployed persons with low education to find work is supported by the experience of three recent ESF-funded programmes from Hungary, although the information on the efficiency of these programmes is questionable. The ’Improvement of employability of the disadvantaged’ (SROP 1.1.2) implemented in 2008-2010 and the ’Road to the world of work’ (SROP 1.1.3) implemented in 2010, involved a customised combination of counselling, mentoring, combined with training programmes and wage subsidies. Both programmes targeted disadvantaged jobseekers, the first, smaller scale (with 6,000 participants) addressed uneducated unemployed on unemployment assistance. The second, larger programme (with 58,000 participants) exhibited a more heterogeneous group. A recent study48_{evaluated the impact of the programme for uneducated} unemployed. Results indicated that participants had a substantially higher probability of being employed (25 percentage points) half a year following the end of the programme than comparable non-participants. Furthermore, the programme was estimated to generate gains (with a cost-to benefit ratio of 1:1.5).49_{The ‘One step ahead!’ (SROP} 2.1.1 and 2.1.1B) programme implemented in 2008-2010 provided a combination of services, including general or vocational training to unemployed with low education and had about 16,500 participants. The evaluation of this programme yielded a large positive impact (30-38 percentage points) of participation on the probability to be employed for uneducated long-term unemployed. Since this programme was less costly than the two mentioned above, it was estimated that it generated a cost-to-gain ratio of 1:2.50

It is worth noting that an individualised complex approach has recently been piloted in the form of coaching and counselling by PES staff for long-term unemployed after they

47_{P.S. Martins and S. Pessoa e Costa (2014) - Reemployment effects from increased activation:} Evidence from times of crisis. IZA DP No. 8600. This study used regression-discontinuity methods, comparing the outcomes of those slightly below the eligibility threshold (having a 6-month long unemployment spell), and those slightly below.

48_{Adamecz et al. (2013), Roma inclusion and impact evaluation of two mainstream EU-funded} active labour market programs. Budapest Institute Working Paper. This study used matching methods to evaluate the programmes.

49_{The cost of the programme was estimated to 920,000 HUF (per participant). Re-employment} entailed an increase to the public budget of 60,000 HUF per month (23,000 HUF in

unemployment assistance saved and 37,000 HUF in tax and social security contributions). While the programme would need 15 months of additional employment to break-even, it likely shortened unemployment durations by about 24 months.

50_{The cost of the programme was estimated to about 780,000 HUF (per participant). It would have} reached a break-even point if employment spells were extended by 13 months. The

(26)

have found employment in a handful of countries.51_{While there is some preliminary} evidence that this approach leads to improvements in (former) job stability for long-term unemployed persons, there are no impact evaluations and cost-benefit results currently available.

Interim summary

In a system of “individual support”, all members of the target group have access to the full spectrum of ALMP-measures. This implies that such measures are available for the target group, but that their contact persons have discretion to decide whether a person is allocated to a measure or not. Hence, it has to be emphasized that ALMP-measures can be an element of labour market oriented service provision, but do not necessarily have to be. ALPM-measures are, however, by no means synonymous with labour market oriented service provision and along the same lines, activation of the target group should by no means be equated with the number of participants in ALMP-measures.

With respect to the provision of ALMP-measures for members of the target group, the available evidence suggests that the allocation process to such measures is extremely important. With particular respect to employment measures on the secondary labour market, an extremely careful selection of participants is necessary to avoid negative effects on their employment prospects. Therefore, such measures are by no means suitable for large-scale use. By contrast, since a large share of factual spending on ALMP-measures is devoted to employment ALMP-measures on the secondary labour market, it seems possible to reduce them and to use the resources to counter-finance a more personnel-intensive guidance and counselling process.

Furthermore, the available evidence indicates that short-term training measures (up to two months) are effective and efficient, especially if they are organized in a company (e.g. as internship). This holds particularly for younger jobseekers. Specialized measures for selected sub-groups, such as clients planning self-employment, also tend to exhibit positive effects. Moreover, complex programmes allowing counsellors to select a combination of existing measures which appear more suited for the target group, also display mixed results regarding effectiveness and efficiency. However, in general such measures tend to be promising.

Finally, it is worth noting that some studies demonstrate considerable effect heterogeneity with respect to the specific contents of measures as well as the providers of such measures. From a practical point of view this implies that it seems to be advisable to establish a system of quality monitoring/management regarding the contents of measures and a selection of service providers according to their effectiveness.

Intervention 4: Provision of a “high-intensity labour-market oriented service bundle” This service bundle includes all elements of the “individual standard support” (see above) but with caseloads of 1:100 and shorter contact intervals (once per month). Hence, it is characterized by intensified counselling and guidance for which lower caseloads for counsellors (i.e. better counsellor-to-client ratios) and shorter contact intervals are the